Stats not counting all work units?
Posted: Sat Sep 06, 2008 12:26 pm
I recently added some extra folding power but I had the feeling that the added power was not reflected in the stats.
I therefore did some calculations myself and it seems that indeed the stats are missing a number of work units.
I would appreciate if someone can take a closer look at what is happening behind the scenes. I don't mind allocating the computer power but only if it used.
However, I do not know what kind of data I should post since it is quiet a lot.
Let me try to explain what I have done (plz note all dates and times are in UTC)
September 5th
-----------------
I switched off all folding activities in order to make sure that no work units were pending at the stats server.
After more than 6 hours I restarted all computers but not at the same time. This should make it easier to debug since work units would be uploaded at different time stamps.
After those 6 hours I had the following stats (consider this the start of my experiment):
2 hours later, I had the same stat values (still normal)
2 hours later, the stats were updated
These stats do reflect the correct number of work units uploaded (5) and the score assigned to it (4 x 480 + 84)
I kept manually tracking the data stored in the log files and stored it in a spreadsheet and the stats were returning the same data as the data in the spreadsheet until September 6th, 01:00.
I then went to bed and this morning I added the remainder of the data to the spreadsheet and found that more data was not reflected in the stats.
According to the stats I upload 50 work units between Sep 5 09:00:03 and Sep 6 11:00:00 and a score increase of 25740.
According to my spreadsheet I uploaded 54 work units and a score increase of 28043.
I have been running this setup for a week now and although FahMon tells me that I should get a score of 28000 PPD, it is more like 22000.
I therefore decided to do the manual calculation and found the same values as reported by FahMon.
All help would be really appreciated
I therefore did some calculations myself and it seems that indeed the stats are missing a number of work units.
I would appreciate if someone can take a closer look at what is happening behind the scenes. I don't mind allocating the computer power but only if it used.
However, I do not know what kind of data I should post since it is quiet a lot.
Let me try to explain what I have done (plz note all dates and times are in UTC)
September 5th
-----------------
I switched off all folding activities in order to make sure that no work units were pending at the stats server.
After more than 6 hours I restarted all computers but not at the same time. This should make it easier to debug since work units would be uploaded at different time stamps.
After those 6 hours I had the following stats (consider this the start of my experiment):
Code: Select all
Fri Sep 5 09:00:03
----------------------
Score: 2299317
WU: 3610
Code: Select all
Fri Sep 5 11:00:00
----------------------
Score: 2299317
WU: 3610
Code: Select all
Fri Sep 5 13:00:01
----------------------
Score: 2301321
WU: 3615
I kept manually tracking the data stored in the log files and stored it in a spreadsheet and the stats were returning the same data as the data in the spreadsheet until September 6th, 01:00.
I then went to bed and this morning I added the remainder of the data to the spreadsheet and found that more data was not reflected in the stats.
According to the stats I upload 50 work units between Sep 5 09:00:03 and Sep 6 11:00:00 and a score increase of 25740.
According to my spreadsheet I uploaded 54 work units and a score increase of 28043.
I have been running this setup for a week now and although FahMon tells me that I should get a score of 28000 PPD, it is more like 22000.
I therefore decided to do the manual calculation and found the same values as reported by FahMon.
All help would be really appreciated